@prof_grimmer

Visiting Assist Prof @MITSloan Assist Prof @JohnsHopkinsAMS, Optimization PhD from @Cornell_ORIE Here to share pretty maths/3D prints/research

 Baltimore, MD
ams.jhu.edu/~grimmer/
Joined January 2015
  • Tweets496
  • Following447
  • Followers3.3K
153Media
Ben Grimmer retweeted
Alex Shtoff
@AlexShtfSep 21
Its been some time since I've read a really nice optimization paper: arxiv.org/abs/2104.11179v5 In short - a beautiful theory for transforming arbitrary constrained problems into equivalent unconstrained ones with a Lipschitz objective.

Radial Duality Part I: Foundations

(Renegar, 2016) introduced a novel approach to transforming generic conic optimization problems into unconstrained, uniformly Lipschitz continuous minimization. We introduce {\it radial...

arxiv.org
13
70
5,230
This week, I am visiting Brown (Monday) and Cornell (Friday), speaking about "Optimization in the Space of Algorithms", Discussing metaquestions about how to enumerate all optimal first-order algs or how to optimally adjust an alg at runtime. If you are in the areas, let me know
1
3
45
2,546
Ben Grimmer retweeted
Emily Riehl@emilyriehlSep 9
According to my memory from last night and the wayback machine, OpenAI updated the Navier-Stokes pdf at Tue, 08 Sep 2026 19:09:35 GMT. The original version was one page shorter and did not contain any citations to Diego Córdoba and Luis Martínez-Zoroa: web.archive.org/web/20260000…
13
134
14
1,310
63,450
Ben Grimmer retweeted
Yuhan Ye@YyuhanYeSep 7
The Silver Rate Is Tight for GD! After non-trivial interaction with ChatGPT 6, we prove the lower bound n^{-(p+C\sqrt{\log\log n/\log n})}, p=\log_2(1+\sqrt{2}), for GD with predetermined nonnegative stepsizes in smooth convex optimization. The silver exponent is optimal.
10
10
2
88
16,828
Wanted to spotlight some very neat math by Kevin Shu and Alex L. Wang: Their recent preprint gives teo explanation of so-called "H-dualities". These are (magical imho) relationships between algorithms where reversing the order of stepsizes used provides new meaningful methods
3
2
1
28
3,815
Typo "teo" -> "two" above is proof my tweets aren't AI generated xD
209
(2) Whenever H dualities hold, a simple algebraic process converts the association "PEP" proof. The analysis behind these insights is pretty slick, working equally for gradient methods and fixed point operators. It's a fun read for anyone interested in mathematical optimization
1
1
258
Link for those interested: arxiv.org/abs/2609.03281

A Unified Theory of H-Duality in First-Order Methods

We provide two complementary explanations of H-duality in smooth strongly convex optimization and contractive fixed-point problems. H-duality refers to the phenomenon where the worst-case...

arxiv.org
2
248
A Unified Theory of H-Duality in First-Order Methods arXiv:2609.03281 On the Complexity of Recognizing SDP Exactness for the Maximum Cut P... arXiv:2609.03508 Beyond Higher-Pulse Rectification: Operational Harmonic Coordination... arXiv:2609.03531
2
2
203
Ben Grimmer retweeted
Emily Riehl@emilyriehlSep 3
Apply for a tenure track job in the Johns Hopkins math department.
David Savitt@dsavittSep 3
We're hiring! We have one tenure-track position this year, in any area of pure math. If you know anyone who might be interested, please let them know! mathjobs.org/jobs/list/28837

MathJobs from the the American Mathematical Society

Mathjobs is an automated job application system sponsored by the AMS.

mathjobs.org
3
33
5,903
Ben Grimmer retweeted
Daniel Litt
@littmathSep 2
there’s a lot of alpha in actually caring about the quality of your work
13
89
8
970
59,815
I quite like this paper and Shuvo's past pre-LLM work automating first-order design. It's a bold new world. I recommend reading and following @shuvo_das_gupta (and @ErnestRyu for good measure)
Shuvomoy Das Gupta
@shuvo_das_guptaAug 18
What role remains for human researchers as frontier LLMs offer remarkable mathematical intelligence? With Heechang Kim and @ErnestRyu, we add a domain-specific harness to general-purpose LLMs, giving them capabilities unavailable on their own: arxiv.org/abs/2608.07407 (1/7)
1
11
1,676
What role remains for human researchers as frontier LLMs offer remarkable mathematical intelligence? With Heechang Kim and @ErnestRyu, we add a domain-specific harness to general-purpose LLMs, giving them capabilities unavailable on their own: arxiv.org/abs/2608.07407 (1/7)
3
9
3
66
24,041
Ben Grimmer retweeted
Jianhao Ma@jianhao_maAug 16
We used GPT-5.6 Sol Pro to prove a new lower bound for gradient descent in smooth convex optimization. For GD with arbitrary predetermined step sizes, we prove \Omega(T^{-1.9319}). arxiv.org/abs/2608.10418

A lower bound for stepsize-based acceleration of gradient descent

Recent work has shown that, for smooth convex optimization, plain gradient descent can be accelerated from its textbook convergence rate of $O(T^{-1})$ (where $T$ denotes the number of iterations)...

arxiv.org
15
56
9
647
76,932
A new paper by Jianhao Ma and Yuxin Chen, with proof "developed by GPT-5.6 Sol Pro", answers a question I have cared about for the past few years. They showed that no gradient descent stepsize schedule (fractally or otherwise) can get full acceleration, i.e., matching Nesterov.
5
25
3
154
19,559
Work from Jason Altschuler and Pablo Parrilo and from myself, Kevin Shu and Alex Wang, showed you can get p=1.2716. This new result shows you cannot get p>1.9319. I strongly believe our p=1.2716 is tight (even in asymp constants), so just another 1.9319-1.2716 = 0.6603 to go :)
1
1
12
2,017
Link to their work for those interested: arxiv.org/abs/2608.10418

A lower bound for stepsize-based acceleration of gradient descent

Recent work has shown that, for smooth convex optimization, plain gradient descent can be accelerated from its textbook convergence rate of $O(T^{-1})$ (where $T$ denotes the number of iterations)...

arxiv.org
2
22
1,876
IIT (my undergrad) just laid off 160 (tenured) faculty and staff. They were certainly more exposed to federal grants and international students than most but I fear this is a sign of more to come Unclear what is left of their/my applied math department... wbez.org/education/2026/07/3…

Illinois Institute of Technology cuts 160 faculty and staff just before fall semester

University officials say finances have taken a hit from declines in research funding and foreign student enrollment. Faculty say leaders’ slow response and lack of transparency have deepened the...

wbez.org
2
14
4
99
198,798
Load more